Hugging Face's logo

Spaces:
metacritical
/
DeepSeekPapers
Running

App Files Community
2
metacritical commited on
Commit
4af19e1
·
verified ·
1 Parent(s): 89f340d

Other style.

Browse files
Files changed (1) hide show
  1. index.html +106 -102
index.html CHANGED
@@ -29,9 +29,9 @@
29
  <div class="container is-max-desktop">
30
  <div class="columns is-centered">
31
  <div class="column has-text-centered">
32
- <h1 class="title is-1 publication-title">DeepSeek Papers</h1>
33
  <div class="is-size-5 publication-authors">
34
- Advancing Open-Source Language Models
35
  </div>
36
  </div>
37
  </div>
@@ -44,12 +44,11 @@
44
  <!-- Abstract. -->
45
  <div class="columns is-centered has-text-centered">
46
  <div class="column is-four-fifths">
47
- <h2 class="title is-3">DeepSeek Research Contributions</h2>
48
  <div class="content has-text-justified">
49
  <p>
50
- Below is a list of significant papers by DeepSeek detailing advancements in large language models (LLMs),
51
- ordered by release date from most recent to oldest. Each paper includes a brief description and highlights
52
- upcoming deep dives.
53
  </p>
54
  </div>
55
  </div>
@@ -57,111 +56,107 @@
57
  <!--/ Abstract. -->
58
 
59
  <!-- Paper Collection -->
60
- <div class="columns is-centered">
61
  <div class="column is-four-fifths">
62
- <div class="content">
63
- <div class="publication-list">
64
- <!-- Papers in chronological order -->
65
- <div class="publication-item">
66
- <div class="publication-title">
67
- <a href="#">DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning</a>
68
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
69
- </div>
70
- <div class="publication-info">
71
- <strong>Release Date:</strong> January 20, 2025
72
- </div>
73
- <div class="publication-description">
74
- The R1 model enhances reasoning capabilities through large-scale reinforcement learning, competing
75
- directly with leading models like OpenAI's o1.
76
- </div>
77
  </div>
 
 
 
 
 
 
78
 
79
- <div class="publication-item">
80
- <div class="publication-title">
81
- <a href="#">DeepSeek-V3 Technical Report</a>
82
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
83
- </div>
84
- <div class="publication-info">
85
- <strong>Release Date:</strong> December 2024
86
- </div>
87
- <div class="publication-description">
88
- This report discusses the scaling of sparse MoE networks to 671 billion parameters, utilizing mixed
89
- precision training and HPC co-design strategies.
90
- </div>
91
  </div>
 
 
 
 
 
 
92
 
93
- <div class="publication-item">
94
- <div class="publication-title">
95
- <a href="#">DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model</a>
96
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
97
- </div>
98
- <div class="publication-info">
99
- <strong>Release Date:</strong> May 2024
100
- </div>
101
- <div class="publication-description">
102
- This paper introduces a Mixture-of-Experts (MoE) architecture, enhancing performance while reducing
103
- training costs by 42%.
104
- </div>
105
  </div>
 
 
 
 
 
106
 
107
- <div class="publication-item">
108
- <div class="publication-title">
109
- <a href="#">DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models</a>
110
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
111
- </div>
112
- <div class="publication-info">
113
- <strong>Release Date:</strong> April 2024
114
- </div>
115
- <div class="publication-description">
116
- This paper presents methods to improve mathematical reasoning in LLMs, introducing the Group
117
- Relative Policy Optimization (GRPO) algorithm.
118
- </div>
119
  </div>
 
 
 
 
 
120
 
121
- <div class="publication-item">
122
- <div class="publication-title">
123
- <a href="#">DeepSeekLLM: Scaling Open-Source Language Models with Longer-termism</a>
124
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
125
- </div>
126
- <div class="publication-info">
127
- <strong>Release Date:</strong> November 29, 2023
128
- </div>
129
- <div class="publication-description">
130
- This foundational paper explores scaling laws and the trade-offs between data and model size,
131
- establishing the groundwork for subsequent models.
132
- </div>
133
  </div>
 
 
 
 
 
134
 
135
- <div class="publication-item">
136
- <div class="publication-title">
137
- <a href="#">DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data</a>
138
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
139
- </div>
140
- <div class="publication-description">
141
- Focuses on enhancing theorem proving capabilities in language models using synthetic data for training.
142
- </div>
143
- </div>
144
 
145
- <div class="publication-item">
146
- <div class="publication-title">
147
- <a href="#">DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence</a>
148
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
149
- </div>
150
- <div class="publication-description">
151
- This paper details advancements in code-related tasks with an emphasis on open-source methodologies,
152
- improving upon earlier coding models.
153
- </div>
154
- </div>
155
 
156
- <div class="publication-item">
157
- <div class="publication-title">
158
- <a href="#">DeepSeekMoE</a>
159
- <span class="tag is-info is-light">[Deep Dive Coming Soon]</span>
160
- </div>
161
- <div class="publication-description">
162
- Discusses the integration and benefits of the Mixture-of-Experts approach within the DeepSeek framework.
163
- </div>
164
- </div>
165
  </div>
166
  </div>
167
  </div>
@@ -172,10 +167,19 @@
172
  <footer class="footer">
173
  <div class="container">
174
  <div class="content has-text-centered">
175
- <p>
176
- This website is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative
177
- Commons Attribution-ShareAlike 4.0 International License</a>.
178
- </p>
 
 
 
 
 
 
 
 
 
179
  </div>
180
  </div>
181
  </footer>
 
29
  <div class="container is-max-desktop">
30
  <div class="columns is-centered">
31
  <div class="column has-text-centered">
32
+ <h1 class="title is-1 publication-title">DeepSeek: Advancing Open-Source Language Models</h1>
33
  <div class="is-size-5 publication-authors">
34
+ A collection of groundbreaking research papers in AI and language models
35
  </div>
36
  </div>
37
  </div>
 
44
  <!-- Abstract. -->
45
  <div class="columns is-centered has-text-centered">
46
  <div class="column is-four-fifths">
47
+ <h2 class="title is-3">Overview</h2>
48
  <div class="content has-text-justified">
49
  <p>
50
+ DeepSeek has released a series of significant papers detailing advancements in large language models (LLMs).
51
+ Each paper represents a step forward in making AI more capable, efficient, and accessible.
 
52
  </p>
53
  </div>
54
  </div>
 
56
  <!--/ Abstract. -->
57
 
58
  <!-- Paper Collection -->
59
+ <div class="columns is-centered has-text-centered">
60
  <div class="column is-four-fifths">
61
+ <h2 class="title is-3">Research Papers</h2>
62
+
63
+ <!-- Paper 1 -->
64
+ <div class="publication-block">
65
+ <div class="publication-header">
66
+ <h3 class="title is-4">DeepSeekLLM: Scaling Open-Source Language Models with Longer-termism</h3>
67
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
68
+ <div class="is-size-5 publication-authors">
69
+ Released: November 29, 2023
 
 
 
 
 
 
70
  </div>
71
+ </div>
72
+ <div class="content has-text-justified">
73
+ <p>This foundational paper explores scaling laws and the trade-offs between data and model size,
74
+ establishing the groundwork for subsequent models.</p>
75
+ </div>
76
+ </div>
77
 
78
+ <!-- Paper 2 -->
79
+ <div class="publication-block">
80
+ <div class="publication-header">
81
+ <h3 class="title is-4">DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model</h3>
82
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
83
+ <div class="is-size-5 publication-authors">
84
+ Released: May 2024
 
 
 
 
 
85
  </div>
86
+ </div>
87
+ <div class="content has-text-justified">
88
+ <p>Introduces a Mixture-of-Experts (MoE) architecture, enhancing performance while reducing
89
+ training costs by 42%.</p>
90
+ </div>
91
+ </div>
92
 
93
+ <!-- Additional papers following same structure -->
94
+ <div class="publication-block">
95
+ <div class="publication-header">
96
+ <h3 class="title is-4">DeepSeek-V3 Technical Report</h3>
97
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
98
+ <div class="is-size-5 publication-authors">
99
+ Released: December 2024
 
 
 
 
 
100
  </div>
101
+ </div>
102
+ <div class="content has-text-justified">
103
+ <p>Discusses the scaling of sparse MoE networks to 671 billion parameters.</p>
104
+ </div>
105
+ </div>
106
 
107
+ <div class="publication-block">
108
+ <div class="publication-header">
109
+ <h3 class="title is-4">DeepSeek-R1: Incentivizing Reasoning Capability in LLMs</h3>
110
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
111
+ <div class="is-size-5 publication-authors">
112
+ Released: January 20, 2025
 
 
 
 
 
 
113
  </div>
114
+ </div>
115
+ <div class="content has-text-justified">
116
+ <p>Enhances reasoning capabilities through large-scale reinforcement learning.</p>
117
+ </div>
118
+ </div>
119
 
120
+ <div class="publication-block">
121
+ <div class="publication-header">
122
+ <h3 class="title is-4">DeepSeekMath: Pushing the Limits of Mathematical Reasoning</h3>
123
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
124
+ <div class="is-size-5 publication-authors">
125
+ Released: April 2024
 
 
 
 
 
 
126
  </div>
127
+ </div>
128
+ <div class="content has-text-justified">
129
+ <p>Presents methods to improve mathematical reasoning in LLMs.</p>
130
+ </div>
131
+ </div>
132
 
133
+ <div class="publication-block">
134
+ <div class="publication-header">
135
+ <h3 class="title is-4">DeepSeek-Prover: Advancing Theorem Proving in LLMs</h3>
136
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
137
+ </div>
138
+ <div class="content has-text-justified">
139
+ <p>Focuses on enhancing theorem proving capabilities using synthetic data for training.</p>
140
+ </div>
141
+ </div>
142
 
143
+ <div class="publication-block">
144
+ <div class="publication-header">
145
+ <h3 class="title is-4">DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models</h3>
146
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
147
+ </div>
148
+ <div class="content has-text-justified">
149
+ <p>Details advancements in code-related tasks with emphasis on open-source methodologies.</p>
150
+ </div>
151
+ </div>
 
152
 
153
+ <div class="publication-block">
154
+ <div class="publication-header">
155
+ <h3 class="title is-4">DeepSeekMoE: Advancing Mixture-of-Experts Architecture</h3>
156
+ <span class="tag is-primary is-medium">Deep Dive Coming Soon</span>
157
+ </div>
158
+ <div class="content has-text-justified">
159
+ <p>Discusses the integration and benefits of the Mixture-of-Experts approach.</p>
 
 
160
  </div>
161
  </div>
162
  </div>
 
167
  <footer class="footer">
168
  <div class="container">
169
  <div class="content has-text-centered">
170
+ <a class="icon-link" href="https://github.com/deepseek-ai" target="_blank" class="external-link">
171
+ <i class="fab fa-github"></i>
172
+ </a>
173
+ </div>
174
+ <div class="columns is-centered">
175
+ <div class="column is-8">
176
+ <div class="content">
177
+ <p>
178
+ This website is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative
179
+ Commons Attribution-ShareAlike 4.0 International License</a>.
180
+ </p>
181
+ </div>
182
+ </div>
183
  </div>
184
  </div>
185
  </footer>